nexarch/dms/internal/batchsplit/render.go

package batchsplit

import (
	"context"
	"fmt"
	"os/exec"
	"path/filepath"
	"regexp"
	"strconv"
	"time"
)

// Renderer rastert die Seiten eines PDFs; Ergebnis: Seitennummer -> PNG-Pfad.
type Renderer interface {
	Render(pdfPath, outDir string) (map[int]string, error)
}

// PopplerRenderer rastert mit pdftoppm (Graustufen-PNG).
type PopplerRenderer struct {
	Binary  string
	DPI     int
	Timeout time.Duration
}

var renderedPage = regexp.MustCompile(`-(\d+)\.png$`)

func (r PopplerRenderer) Render(pdfPath, outDir string) (map[int]string, error) {
	bin, dpi, timeout := r.Binary, r.DPI, r.Timeout
	if bin == "" {
		bin = "pdftoppm"
	}
	if dpi <= 0 {
		dpi = 200
	}
	if timeout <= 0 {
		timeout = 3 * time.Minute
	}
	ctx, cancel := context.WithTimeout(context.Background(), timeout)
	defer cancel()
	prefix := filepath.Join(outDir, "page")
	if out, err := exec.CommandContext(ctx, bin, "-r", strconv.Itoa(dpi), "-gray", "-png", pdfPath, prefix).CombinedOutput(); err != nil {
		return nil, fmt.Errorf("%s: %w: %.200s", bin, err, out)
	}
	files, err := filepath.Glob(prefix + "-*.png")
	if err != nil {
		return nil, err
	}
	pages := make(map[int]string, len(files))
	for _, f := range files {
		if m := renderedPage.FindStringSubmatch(f); m != nil {
			n, _ := strconv.Atoi(m[1])
			pages[n] = f
		}
	}
	return pages, nil
}